Asymptotic linear expansion of profile likelihood in the Cox mode
Y. Hirose · Mathematical Methods of Statistics · 2011
By introducing a new parameterization, Hirose [12] improved on the seminal work of Murphy and van der Vaart [16]: the improvement establishes the efficiency of the estimator through direct quadratic expansion of the profile likelihood, which requires fewer assumptions. This paper aims to demonstrate that the approach in [12] is fully applicable to the Cox proportional hazard model.